AI-Informed Solvation Engineering for Thermogalvanic Electrolytes with High Thermopower.

Area of Science:

• Materials Science
• Electrochemistry
• Computational Chemistry

Background:

• Thermogalvanic cells convert heat to electricity using redox couples.
• Electrolyte performance is hindered by complex ion solvation and thermopower relationships.
• Developing high-performance solvents is crucial for efficient energy conversion.

Purpose of the Study:

• To present a data-driven machine learning strategy for identifying high-performance solvents for thermogalvanic electrolytes.
• To establish a correlation between DFT-derived features and the Seebeck coefficient.
• To accelerate the discovery of novel electrolytes for thermogalvanic applications.

Main Methods:

• A hybrid machine learning framework combining high-throughput DFT calculations and interval regression.
• Parametrization of the Seebeck coefficient using ion-solvent interaction features.
• SHAP analysis to identify dominant features influencing thermopower.
• Experimental validation and model augmentation with solvation-relevant molecular features.

Main Results:

• The ML framework successfully screened extensive solvent candidates.
• Polarity-related features were identified as dominant factors for thermopower.
• The approach validated known effective solvents and identified new promising candidates.
• Experimental validation confirmed the performance of newly identified solvents.

Conclusions:

• The solvation-driven ML approach significantly accelerates electrolyte discovery for thermogalvanic cells.
• The study provides fundamental insights into designing electrolytes with large solvation entropy differences.
• This methodology offers a pathway to optimize heat-to-electricity conversion efficiency.